AI Usage Policy

HEMA AI INC.

Artificial Intelligence Usage Policy

& Editorial Standards

Effective Date: June 13, 2026

Last Updated: June 13, 2026

Maintained by Hema AI Inc. | A Delaware Corporation

Table of Contents

1. Purpose and Scope

2. Definitions

3. Radical Transparency Principles

4. Permitted Uses of Generative AI

5. Prohibited Uses of Generative AI

6. Human Governance and Editorial Review

7. Disclosure Standards for Contributors and Employees

8. Data Privacy, Security, and Confidentiality

9. Intellectual Property and Copyright Compliance

10. Accuracy, Fact-Checking, and Source Attribution

11. Sensitive Topic Protocols

12. Recordkeeping and Audit Trail

13. Training, Onboarding, and Compliance

14. Policy Violations and Enforcement

15. Amendments and Version Control

16. Appeals, Feedback, and Contact

Policy Summary at a Glance

1. Purpose and Scope

1.1 Purpose. This Artificial Intelligence Usage Policy and Editorial Standards (this "Policy") is adopted by Hema AI Inc. ("Company," "we," "our," or "us"), a corporation incorporated under the laws of the State of Delaware, United States of America, with its registered office at 131 Continental Dr, Suite 305, Newark, Delaware 19713. This Policy establishes the standards, obligations, and procedures governing the use of generative artificial intelligence and related automated systems by the Company, its employees, contractors, contributors, agents, and any other individuals acting on behalf of the Company (collectively, "Covered Persons").

1.2 Scope. This Policy applies to all AI-assisted activities undertaken by Covered Persons in connection with the Company's business, including but not limited to: content creation and publication, product development, client communications, research and analysis, marketing, data processing, and internal operations. This Policy applies to all generative AI tools, large language models, image generation systems, and other AI-assisted automation tools used by Covered Persons, whether proprietary to the Company or provided by third parties.

1.3 Governing Law. This Policy is governed by and shall be interpreted in accordance with the laws of the State of Delaware and the federal laws of the United States of America. The Company operates in compliance with applicable U.S. federal and state regulations, including but not limited to those issued by the Federal Trade Commission (FTC), the U.S. Copyright Office, applicable state consumer protection statutes, and emerging AI-specific federal guidance.

1.4 Relationship to Other Policies. This Policy supplements and should be read in conjunction with the Company's Privacy Policy, Terms of Use, Employee Handbook, and any applicable client contracts or data processing agreements. In the event of any conflict between this Policy and another Company policy with respect to AI usage, this Policy shall govern unless a more specific agreement or policy expressly provides otherwise.

2. Definitions

For purposes of this Policy, the following terms shall have the meanings set forth below:

• "Artificial Intelligence" or "AI" means machine-based systems that can generate outputs such as text, images, audio, video, predictions, recommendations, or decisions based on inputs, using machine learning, deep learning, large language models, or similar techniques.

• "Generative AI" means AI systems capable of generating novel content, including text, images, code, audio, and other media, in response to prompts or instructions provided by a user. Examples include, without limitation, large language models such as OpenAI's GPT series, Google's Gemini series, Anthropic's Claude series, and image generation systems such as Midjourney and DALL-E.

• "AI-Assisted Content" means any content, output, communication, or work product in which generative AI materially contributed to the ideation, drafting, structuring, analysis, or refinement of the content, regardless of the degree of subsequent human editing.

• "Covered Person" means any employee, contractor, consultant, advisor, contributor, intern, or other individual who performs work for or on behalf of Hema AI Inc., whether or not such individual is a party to a formal employment or services agreement.

• "Responsible Editor" means the designated human editorial personnel assigned to review, verify, approve, and assume accountability for any AI-Assisted Content prior to its publication or delivery.

• "Personal Information" has the meaning ascribed to it in the Company's Privacy Policy and applicable U.S. data protection laws, including any information that identifies or is reasonably capable of identifying a natural person.

• "Sensitive Topics" means subject matter that involves medical, legal, financial, psychological, political, or other guidance that could foreseeably affect the health, safety, finances, legal rights, or wellbeing of any individual or class of individuals.

3. Radical Transparency Principles

Radical transparency is the Company's foundational operating philosophy with respect to AI usage: we disclose exactly how our work is done, attribute the tooling, and surface evidence that supports every conclusion we publish or deliver. The following principles govern how AI is integrated into all Hema AI content and operations:

3.1 Origin Clarity

Users, clients, and the public shall always have access to information identifying who initiated an idea, drafted a narrative or output, and approved the final publication or delivery. Where AI materially contributed to the origination of an idea or the structure of an output, that contribution shall be disclosed in accordance with Section 7 of this Policy.

3.2 Tool Attribution

When AI shapes ideation, analysis, or phrasing in any material respect, the Company shall identify the specific system or model employed (e.g., Google Gemini Pro 2.5, OpenAI GPT-4.1, Anthropic Claude Sonnet) and describe the nature of its role in the production of the content. Generic or vague attributions (e.g., 'AI-assisted') are insufficient unless accompanied by a specific description of the AI's contribution.

3.3 Evidence Over Opinion

No assertion, statistic, data point, or factual claim shall be published or delivered without a primary source, dataset, official filing, or reproducible methodology that the recipient can independently inspect and verify. AI-generated citations or references shall not be treated as reliable without independent human verification against primary sources.

3.4 Version Control and Audit Trail

Significant updates to AI-Assisted Content are to be dated, explained in plain language, and, where practicable, preserved in a change log visible to the intended audience. Internal editorial logs shall record each revision, the identity of the revising party, and the date and nature of the change. See Section 12 for recordkeeping requirements.

3.5 Accountability

The Company retains full legal and ethical accountability for the accuracy, originality, regulatory compliance, and ethical quality of all AI-Assisted Content. Disclosure of AI involvement does not limit, mitigate, or transfer the Company's liability to any third party, including any AI system vendor or developer.

4. Permitted Uses of Generative AI

Subject to the conditions, limitations, and human oversight requirements set forth in this Policy, Covered Persons are authorized to use generative AI tools for the following purposes:

4.1 Research Acceleration

Generative AI may be used to aggregate prior statements, synthesize publicly available information, identify relevant data sources, and accelerate background research. All AI-generated research outputs shall be treated as unverified leads requiring human expert validation and shall not be cited or relied upon without independent verification against primary sources. The Company shall maintain logs of AI-assisted research sessions where the outputs materially inform published content.

4.2 Drafting and Content Refinement

Covered Persons may use generative AI to propose outlines, suggest structures, tighten prose, generate initial drafts, or explore counterarguments and alternative perspectives. Final narratives, analyses, and content must be re-written, substantively edited, contextualized, and approved by qualified human personnel to ensure original voice, factual accuracy, and contextual integrity. No AI-generated draft shall be published or delivered without human re-authorship and editorial review.

4.3 Structured Data, Schema, and Technical Outputs

AI assistants may be used to generate schema markup, JSON structures, code snippets, data visualizations, and other technical outputs. Human reviewers with appropriate technical expertise must confirm the accuracy, security, and fitness for purpose of all AI-generated technical outputs before deployment or delivery to any client or end user.

4.4 Compliance and Quality Monitoring

Internal AI tools may be used to flag potential policy violations, missing disclosures, unverified claims, or content quality issues. Every alert or flag generated by an AI compliance tool shall be reviewed and resolved by a qualified human editor or compliance officer before the relevant content is published, delivered, or acted upon.

4.5 Internal Operations and Productivity

Generative AI may be used for internal operational tasks such as summarizing internal documents, drafting internal communications, scheduling assistance, and administrative task automation, provided that no Confidential Information, Personal Information, or client data is inputted into third-party AI systems in contravention of Section 8 of this Policy.

5. Prohibited Uses of Generative AI

Notwithstanding any other provision of this Policy, the following uses of generative AI are strictly prohibited and shall constitute a material violation of this Policy:

1. Autonomous Publication. No generative AI system shall publish, post, transmit, or otherwise deliver content, data, or communications to any external audience, platform, or client without prior human review and explicit human approval.

2. Autonomous Approval or Decision-Making. Generative AI shall not be used to approve interviews, adjudicate product claims, modify pricing, issue legal opinions, make hiring or termination decisions, or approve financial transactions, contracts, or commitments on behalf of the Company.

3. Autonomous Code Deployment. Generative AI shall not independently push, deploy, or commit code to any production environment, client-facing system, or live platform without explicit human review, testing, and authorization.

4. Inputting Confidential or Protected Information. Covered Persons shall not input Personal Information, confidential client data, proprietary trade secrets, non-public financial data, legally privileged information, or other protected information into any third-party generative AI system without a valid contractual and legal basis, including appropriate data processing agreements and security safeguards. See Section 8.

5. Fabrication or Misrepresentation. Covered Persons shall not knowingly use AI to fabricate, falsify, or materially misrepresent facts, statistics, quotations, citations, endorsements, or any other information, or to create deceptive or misleading content.

6. Deceptive Attribution. Content generated primarily by AI shall not be represented to any audience or recipient as having been authored, researched, or independently verified by a human expert unless such authorship, research, or verification has in fact occurred.

7. Copyright or IP Infringement. Covered Persons shall not use generative AI to reproduce, adapt, or distribute copyrighted works of third parties without a valid license or legal basis. See Section 9.

8. Discrimination and Harmful Bias. Generative AI shall not be used in any manner that generates, amplifies, or perpetuates discrimination, unlawful bias, harassment, or hate speech on the basis of race, color, religion, sex, national origin, age, disability, sexual orientation, gender identity, or any other characteristic protected under applicable federal or state law.

9. Circumvention of Legal Obligations. Generative AI shall not be used to circumvent, evade, or otherwise avoid compliance with any applicable federal or state law, regulation, court order, or contractual obligation.

6. Human Governance and Editorial Review

6.1 Responsible Editor Requirement. Each item of AI-Assisted Content intended for external publication or client delivery shall be assigned to a designated Responsible Editor. The Responsible Editor shall confirm sourcing integrity, factual accuracy, regulatory compliance (including privacy and copyright compliance), and adherence to this Policy before authorizing publication or delivery.

6.2 Sensitive Topic Review. Content addressing Sensitive Topics — including but not limited to medical information, legal guidance, financial advice, psychological or mental health content, or content targeted at children — requires additional review by a qualified subject-matter expert and, where the content could foreseeably create legal or regulatory exposure for the Company or its users, legal sign-off from qualified legal counsel before publication or delivery.

6.3 Editorial Logs. The Company shall maintain editorial logs for all AI-Assisted Content that is published externally or delivered to clients. Each log entry shall record: (a) the identity of the Responsible Editor; (b) the AI tools that assisted in the production of the content and the nature of their contribution; (c) the date and time of final editorial approval; (d) any changes made during editorial review; and (e) the identity of any subject-matter expert who reviewed the content.

6.4 Corrections. Where published content is found to contain an error, inaccuracy, or misleading statement — whether or not AI-assisted — a correction shall be issued promptly. Corrections shall be visibly noted at the top or bottom of the relevant page or document, with a clear description of what was corrected and the date of correction. Corrections to client-facing deliverables shall also be communicated directly to the affected client.

6.5 No Autonomous Action. Generative AI is deployed at Hema AI as a co-pilot that qualified human experts interrogate, challenge, and verify — it is not an oracle whose outputs the Company defers to. No action, publication, deployment, or decision that could have material consequences for the Company, its clients, or its users shall be taken on the basis of AI outputs alone.

7. Disclosure Standards for Contributors and Employees

7.1 Mandatory Disclosure. All Covered Persons who produce AI-Assisted Content for external publication or client delivery must include a clear and conspicuous disclosure at or near the point of consumption — on-page, inline, and near the author byline where applicable. Disclosure must be made in plain English and must not be buried in footnotes, endnotes, or general terms and conditions in a manner that a reasonable reader would be unlikely to notice.

7.2 Required Disclosure Elements. Each disclosure must, at minimum:

• Identify which specific parts of the content benefited from AI assistance (e.g., research aggregation, outline generation, initial drafting, data visualization, code generation, or copy editing);

• Name the primary AI tools or models used (e.g., Google Gemini Pro 2.5, OpenAI GPT-4.1, Anthropic Claude Sonnet, or proprietary Hema AI internal agents);

• Affirm that a qualified human reviewer has reviewed the content for accuracy, originality, and compliance with this Policy; and

• Include a reference or hyperlink to this Policy (www.tryhema.com/ai-policy) so that readers can review the Company's full AI standards.

7.3 No Retroactive Attribution. Disclosure obligations apply at the time of original publication. If previously published content is materially updated with AI assistance, the updated disclosure must be added or updated at the time the update is made, and the update must be noted in the content's change log.

7.4 Guest Contributors and Advisors. Third-party guest contributors and advisors who submit content for publication on behalf of or in association with Hema AI are required to comply with the disclosure standards set forth in this Section 7 as a condition of publication. The Company reserves the right to reject, edit, or require amendment of any contributed content that does not comply with this Policy.

8. Data Privacy, Security, and Confidentiality

8.1 Compliance with Privacy Policy. All AI-assisted activities undertaken by Covered Persons must comply with the Company's Privacy Policy, applicable U.S. federal and state privacy laws (including but not limited to the California Consumer Privacy Act (CCPA) as amended by the California Privacy Rights Act (CPRA), and applicable sectoral laws), and any contractual data protection obligations owed to clients or users.

8.2 Prohibition on Unauthorized Data Input. Covered Persons are prohibited from inputting any of the following categories of information into any third-party generative AI system without explicit written authorization from the Company's designated privacy officer and, where applicable, a valid data processing agreement with the AI system vendor: (a) Personal Information of any user, client, employee, or third party; (b) non-public financial information, including the Company's financial data or client financial data; (c) legally privileged communications; (d) confidential business information, trade secrets, or proprietary technology; or (e) any information subject to a contractual confidentiality obligation.

8.3 Vendor Due Diligence. Before deploying any third-party AI tool in connection with the Company's operations, the relevant Covered Person or procurement team shall conduct reasonable due diligence to confirm that the vendor's data handling practices are consistent with applicable law and the Company's contractual obligations, including confirming whether the vendor uses submitted inputs for model training and whether appropriate data processing agreements are in place.

8.4 Incident Reporting. Any Covered Person who becomes aware of an actual or suspected unauthorized disclosure of Confidential Information or Personal Information through an AI system shall report the incident to the Company's privacy officer at info@tryhema.com within twenty-four (24) hours of becoming aware of the incident. The Company will investigate and respond to all reported incidents in accordance with its data breach response procedures and applicable law.

9. Intellectual Property and Copyright Compliance

9.1 Ownership of AI-Assisted Outputs. The Company claims ownership of all AI-Assisted Content produced by Covered Persons within the scope of their engagement with the Company, to the extent such content is legally capable of ownership under applicable U.S. copyright law. Covered Persons acknowledge that, as of the effective date of this Policy, the U.S. Copyright Office does not extend copyright protection to content generated solely by AI without sufficient human authorship. Accordingly, Covered Persons are expected to make sufficient original human creative contributions to any AI-Assisted Content to support a valid copyright claim by the Company.

9.2 Third-Party Copyright. Covered Persons must not use generative AI to reproduce, closely paraphrase, or create derivative works of third-party copyrighted materials without a valid license, express permission, or a legally sound fair use basis under 17 U.S.C. § 107. Covered Persons are responsible for ensuring that AI-generated outputs do not infringe third-party intellectual property rights, and must not publish or deliver any AI-assisted output without conducting reasonable verification that the output does not reproduce protected third-party content.

9.3 No Reverse-Engineering of Third-Party Systems. Covered Persons shall not use the Company's AI tools or systems to reverse engineer, circumvent, extract model weights from, or otherwise violate the terms of service of any third-party AI provider.

10. Accuracy, Fact-Checking, and Source Attribution

10.1 Verification Standard. No claim, statistic, data point, quotation, or citation included in any AI-Assisted Content shall be published or delivered without independent human verification against a primary source. Verification must be performed by a Covered Person with appropriate subject-matter expertise. AI-generated citations, references, and data points are to be treated as unverified leads until confirmed by a qualified human reviewer.

10.2 Source Documentation. For all externally published AI-Assisted Content, the Responsible Editor shall maintain a source-of-truth document or log identifying the primary sources relied upon for all material factual claims. This documentation shall be retained for a minimum of three (3) years from the date of publication.

10.3 Prohibition on AI-Generated Hallucinations. Covered Persons are aware that generative AI systems are capable of generating plausible-sounding but factually incorrect information (commonly referred to as 'hallucinations'). The Company strictly prohibits the publication or delivery of any content containing AI-generated information that has not been independently verified. A Covered Person who knowingly publishes unverified AI-generated information shall be subject to disciplinary action under Section 14.

11. Sensitive Topic Protocols

11.1 Identification of Sensitive Topics. Prior to publishing or delivering any AI-Assisted Content that addresses a Sensitive Topic (as defined in Section 2), the Responsible Editor shall identify the sensitive nature of the content and escalate it for specialized review in accordance with this Section 11.

11.2 Expert Review Requirement. AI-Assisted Content addressing Sensitive Topics requires review and written sign-off by a qualified subject-matter expert (e.g., a licensed physician for medical content, a licensed attorney for legal content, or a registered financial professional for financial advice content) before publication or delivery. The identity of the reviewing expert and the date of sign-off shall be recorded in the editorial log.

11.3 Legal Review Requirement. Where AI-Assisted Content addressing a Sensitive Topic could foreseeably create legal, regulatory, or reputational exposure for the Company, its clients, or its users, the content shall be reviewed by qualified legal counsel before publication or delivery.

11.4 Disclaimers. All AI-Assisted Content that addresses medical, legal, financial, or similar Sensitive Topics must include a prominent disclaimer advising readers that the content is for informational purposes only, does not constitute professional advice, and should not be relied upon as a substitute for consultation with a licensed professional.

12. Recordkeeping and Audit Trail

12.1 Editorial Logs. The Company shall maintain written editorial logs for all externally published or client-delivered AI-Assisted Content. Each log shall record, at minimum: (a) the title or description of the content; (b) the date of publication or delivery; (c) the identity of the Responsible Editor and any contributing Covered Persons; (d) the AI tools used and the nature of their contribution; (e) the primary sources verified; (f) the date and identity of any subject-matter expert reviewers; and (g) the date and description of any post-publication corrections or updates.

12.2 Retention. Editorial logs, source documentation, and AI disclosure records shall be retained for a minimum of three (3) years from the date of creation. Financial records relating to AI tool subscriptions and vendor contracts shall be retained in accordance with the Company's general financial recordkeeping obligations under applicable federal and Delaware state law.

12.3 Access and Audit. Editorial logs and AI disclosure records shall be made available to authorized Company personnel, qualified auditors, and, to the extent required by applicable law or regulation, governmental or regulatory authorities. The Company reserves the right to conduct periodic internal audits of AI usage and editorial practices to assess compliance with this Policy.

13. Training, Onboarding, and Compliance

13.1 Mandatory Training. All Covered Persons whose roles involve the use of generative AI tools in connection with the Company's business shall complete mandatory AI usage and ethics training prior to using such tools. Training shall cover the requirements of this Policy, applicable legal obligations, responsible AI practices, and the Company's disclosure standards.

13.2 Onboarding. Compliance with this Policy shall be a condition of engagement for all new employees, contractors, and contributors. Covered Persons shall acknowledge receipt and understanding of this Policy in writing as part of their onboarding process.

13.3 Ongoing Training. The Company shall provide updated training to Covered Persons whenever this Policy is materially amended or whenever significant changes in applicable law, regulation, or AI technology practice necessitate updated guidance.

14. Policy Violations and Enforcement

14.1 Consequences of Violation. Violations of this Policy by employees may result in disciplinary action up to and including termination of employment, in accordance with the Company's Employee Handbook and applicable law. Violations by contractors or contributors may result in immediate termination of the applicable services agreement and, where warranted, legal action to recover damages.

14.2 Reporting Violations. Covered Persons who become aware of an actual or suspected violation of this Policy are encouraged and, in certain circumstances, required to report such violations promptly to the Company's designated compliance contact at info@tryhema.com. Reports may be made confidentially to the extent practicable and permitted by applicable law.

14.3 Non-Retaliation. The Company strictly prohibits retaliation against any Covered Person who reports a suspected policy violation in good faith. Any Covered Person who retaliates against a good-faith reporter shall themselves be subject to disciplinary action up to and including termination.

14.4 FTC Compliance. The Company recognizes that the Federal Trade Commission (FTC) has authority to regulate deceptive AI-related practices as unfair or deceptive acts or practices under Section 5 of the FTC Act (15 U.S.C. § 45). The Company is committed to complying with all applicable FTC guidance and enforcement actions concerning AI usage, disclosure, and transparency.

15. Amendments and Version Control

15.1 Right to Amend. The Company reserves the right to amend, update, or replace this Policy at any time to reflect changes in applicable law, regulatory guidance, AI technology, or Company practices. Amended versions of this Policy will be posted at www.tryhema.com/ai-policy and will identify the date of the most recent update.

15.2 Material Amendments. If a proposed amendment would materially alter the obligations or rights of Covered Persons under this Policy, the Company will provide reasonable advance notice of such amendment to affected Covered Persons before the amendment takes effect.

15.3 Version History. The Company maintains a version history of this Policy. Upon written request to info@tryhema.com, the Company will make prior versions of this Policy available to any Covered Person who requests them.

16. Appeals, Feedback, and Contact

16.1 Content Disputes. Any user, client, or member of the public who believes that content published or delivered by Hema AI Inc. breaches this Policy, contains an inaccuracy, or otherwise fails to meet the Company's editorial standards is invited to contact the Company's editorial team at the address below. The Company will review every submission within two (2) U.S. business days and will respond in writing.

16.2 Legal and Regulatory Inquiries. For legal inquiries, regulatory requests, subpoenas, or other formal legal process, please address correspondence to the Company's registered address set forth below. Service of legal process on the Company shall be made in accordance with the Delaware General Corporation Law and applicable federal procedural rules.

16.3 Contact Information. All inquiries under this Policy should be directed to:

Hema AI Inc.

131 Continental Dr, Suite 305

Newark, Delaware 19713

United States of America

Email: info@tryhema.com

Website: www.tryhema.com

Delaware File Number: 10586210

* * *

Hema AI Inc. | 131 Continental Dr, Suite 305, Newark, Delaware 19713 | info@tryhema.com | www.tryhema.com